ACQUISITION/PROCESSING Coordinated by SHUKI RONEN

Comparing the MEMS accelerometer and the analog geophone

JIDONG WEI, Sinopec

O ver the last decade or two, digital micro-electro-

mechanical system (MEMS) accelerometers have been
ground shaking, and the comparison of the noise levels
again, should be done in the same motion domain, either
Downloaded 11/11/13 to 130.238.140.107. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

proposed as a replacement for analog electromagnetic coiled velocity or acceleration. Figure 2 compares the results
geophones. Although many positive results have been reported from a geophone and an accelerometer, and shows that the
in articles and at conferences, other authors have claimed that DSU3 has lower instrument noise in the high-frequency
there are no obvious differences between them. This article end of the spectrum.
compares the analog geophone and MEMS accelerometer • Certainly, the position of the intersection point for two
and our results show that the MEMS accelerometer has made curves (Figure 2) depends on the sensitivities of the geo-
some improvements on electric specifications. However, these phone and the accelerometer. At the same time, because
improvements are so slight that identifying the enhanced acceleration-type sensors have higher sensitivities, they can
signals obtained by the MEMS accelerometer is extremely almost reach the maximum span of dynamic range. The
challenging, if not impossible. Whether these improvements geophone has lower sensitivity; so, with the same vibra-
can evolve into significant acquisition benefits depends on tion, the output voltage is lower and consequently less dy-
many other factors involved in a seismic survey and which namic range can be used than with its digital counterpart
may contaminate the weak signal and effectively cancel most (Figure 3) even though the accelerometer’s total dynamic
of the accelerometer’s improvement. range is a little less than the geophone’s (Table 1). How-
ever, connecting geophones in series or raising the sensi-
Introduction tivity of the geophone increases the output voltage and
The coil-based geophone is a proven technology that has long will make full use of the AD converter’s dynamic range as
provided the industry with rugged, cheap, and self-powered the DSU does, by which the intersection point shifts right
sensors. However, more quantitative seismic surveys will re- (Figure 2, black curve), implying that the geophone may
quire broader bandwidth and better-calibrated sensors. The have lower instrument noise in the low-frequency end.
accelerometer is one proposed solution. So what are the dif- • The essential benefit of the accelerometer is a broadband
ferences in specifications and dimensions between the accel- linear response of amplitude and phase from 0 (DC) to
erometer and the geophone? 800 Hz. The MEMS resonant frequency (1 kHz) is far
above the seismic band, which makes it possible to record
• The coiled geophone is a velocity-type sensor and most frequencies below 10 Hz without attenuation, even the
accelerometers, such as the DSU3, are acceleration-type. direct current. However, because of the effect of the natu-
The acceleration spectral ratio is proportional to frequen- ral frequency and damping, attenuation of 12 dB/octave
cy and there is a 90° phase shift compared with velocity. happens to a geophone. At the same time, the geophone’s
Therefore, for the same vibration,
the accelerometer’s dominant fre-
quency is higher than that of the
geophone (Figure 1) and, in order
for these two types of sensors to
be meaningfully compared, the
recorded data must be viewed in
the same motion domain (i.e., ve-
locity or acceleration).
• Digitization of the recorded data
is accomplished by different ele-
ments for the geophone and the
accelerometer; the geophone uses
an AD converter and the acceler-
ometer an application-specific in-
tegrated chip (ASIC). Therefore,
the instrument noise is not the
same, and the dynamic ranges are
not identical (Table 1). The dy-
namic range of the accelerometer
(DSU3) is a little more restricted
than that of the geophone (FDU).
• In actual field work, the instru- Figure 1. The dominant frequency of accelerometer/acceleration is higher than that of geophone/
ment noise is regarded as part of velocity.

1206 The Leading Edge October 2013

 ACQUISITION/PROCESSING

spurious frequency of 200–300 Hz (Figure 4) limits the In Figure 6, the ambient noise stacked 600 times could be
geophone’s linear frequency range of working, although regarded as the same as the noise level after processing and any
the geophone’s spurious frequency has little influence signal whose strength is below it could not be identified. (Of
when the reflection’s frequency is low, or when the noise, course, it does not include source-oriented noise.)
both ambient and source-oriented, is strong. The effective dynamic range in Figure 6 is calculated in two
Downloaded 11/11/13 to 130.238.140.107. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

cases: when distortion of a certain reflection > 3 in Figure 6, the

As a consequence of these differences between the two types effective dynamic range = –1 × distortion; when distortion of a
of sensors, two ways to compare them are: filter the accel- certain reflection < 3 in Figure 6, the effective dynamic range =
erometer’s data as a geophone does, or compensate for the 20 × log 10 (maximum value of reflection/3).
low-frequency component loss by a geophone by spike de-
convolution. Response to outside interference
In some earlier comparisons, a 5-Hz low-cut filter was used in Noise from electromagnetic interferences, signal crosstalk,
the accelerometer’s data processing to match the geophone’s filter and sensitivity to leakage was common for transmission be-
response. However, we found that the above two kinds of filters tween strings of geophones and the recording truck. In re-
may have a slight difference in waveform which makes the ac- cent years, the electronics distributed in the seismic network
celerometer’s data appear to have stronger amplitudes (Figure 5). provide digitalization close to the geophone groups. Only
Linear distortion
The accelerometer (DSU3) AD Convert Device (Sample Interval) Floor Noise Dynamic Range
has a low –90 dB harmon- (electric)~Maximum Voltage
ic distortion (it is –60 dB (gain = 0 dB)
for most geophones) but if 428XL-DSU3-ASIC (4ms) 1.8 μv~2262 mv 122 dB
the influence of strong and 428XL-DSU3-ASIC (1ms) 3.6 μv~2262 mv 116 dB
variable noise, unstable cou- 428XL-FDU-AD CONVERT (4ms) 0.45 μv~2262 mv 134 dB
pling effects, and electric
noise greatly exceeds the dis- 428XL-FDU-AD CONVERT (1ms) 0.9 μv~2262 mv 128 dB
tortion, the benefit will com- Table 1. Different AD converter devices and their corresponding dynamic ranges (SERCEL 428 Series, for
pletely disappear (Figure 6). example).

Figure 3. Actual dynamic ranges of two kinds of AD convert devices

Figure 2. Comparison of instrument noise in acceleration domain. (SERCEL 428 Series for example. Gain 0 dB, geophone sensitivity:
(SERCEL 428 Series, for example. Gain 0 dB, geophone sensitivity: 20.1 v/m/s, accelerometer sensitivity: 452 mv/m/s2.)
20.1 v/m/s, accelerometer sensitivity: 452 v/m/s2.)

Figure 4. Different frequency responses. Figure 5. Comparison of accelerometer and geophone in acceleration
domain.

October 2013 The Leading Edge 1207

 ACQUISITION/PROCESSING

two pairs of wires are necessary for telemetric transmission The coupling effect
of thousands of channels in real time. Sensitivity to leakage Materials and the shape and mass of a sensor (which are dif-
is reduced and the digital data are controlled by each node ferent between the geophone and accelerometer) are the main
of the seismic network. However, noise may still contami- factors determining the coupling effect between the sensor
nate the analog signals transmitted through the strings of and the ground. We found that the DSU3 has a different cou-
Downloaded 11/11/13 to 130.238.140.107. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

geophones. This contamination could be eliminated with the pling effect than the geophone and exerts a negative influence
integration of the sensor with the station as is done with ac- on the high-frequency end, usually more than 150 Hz.
celerometers and which reduces noise a great deal. But in A test that showed the influence of the coupling effect
some practical situations, a little EM interference remains. consisted of 36 accelerometers (DSU3, 10 cm apart) and a
The possible reasons are that there are still analog electric shot with an offset of 50 m (Figure 7). The waveforms and
elements in MEMS which could be affected by outside EM spectra of this test (Figure 8) reveal that some differences oc-
interferences, or the acoustic and mechanical vibration com- curred when sensors are planted into the same coupling me-
ing from the air and ground will initiate sensor shaking. dium and in a small and limited area (compared with the
scale in petroleum exploration, tens or several meters). The
Single point or array? differences at the high-frequency end are far above the sen-
Single-point receiving is considered a great advantage for sor’s distortion (–60 dB–90 dB) and make identifying the
the accelerometer because of its fully digital transmission faint signal more difficult. The only explanation for these dif-
but geophones can also benefit from this technique. The ferences is the coupling effect. The coupling effect can pro-
factor which makes the high-frequency component attenu- duce unexpected changes to a weak but still detectable signal.
ate is the array filtering, no matter what kind of sensor is
used. Furthermore, the comparison between geophone and A field case
accelerometer in field operation should be done in a “one- A field test in Jiyang Basin, China, was performed to com-
accelerometer to one-geophone” pattern (instead of “one- pare differences between geophones and accelerometers. The
accelerometer to geophone-group”) because the array’s filter geometry of the test (Figure 9) was 5-m crossline spacing,
response can result in a misleading conclusion. 20-m inline spacing, one 20dx to one DSU3 separated by 20
cm, a source line with 60 shots with
a shot interval of 50 m.
Data were subsequently extracted
from 60 shots of a particular geophone
or accelerometer, producing new
60-trace gathers which imitated shot
gathers of 60 traces. In these gathers,
the only difference was the receiving
device, either geophone or accelerom-
eter. The comparison was done in two
ways: one geophone to one accelerom-
eter or 36 geophones to 36 accelerom-
eters. Figure 10, which compares one
Figure 6. Effective dynamic ranges for geophone and accelerometer.

Figure 7. Geometry of test 1. Figure 8. Waveforms and spectra of different sensors in Figure 7.

1208 The Leading Edge October 2013

 ACQUISITION/PROCESSING

geophone and one accelerometer, shows they were similar both in

waveforms and in amplitude-frequency curves (Figure 8) in the
acceleration domain. The accelerometer data were filtered by a
quasi-geophone filter; the analog data were the differential of the
raw data. Some differences in Figure 7 may come from:
Downloaded 11/11/13 to 130.238.140.107. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

1) The coupling effect. The influence of coupling effect is

much stronger than it is often imagined and most often
affects the high-frequency end.
2) Instrument noise. The instrument noise in voltage of DSU3
is higher than that of FDU, however the DSU3 is an accel-
eration-type sensor and its high-frequency-end instrument
noise is lower than that of geophone in motion domain.
3) The burn-in of sensors, especially the sensitivity, makes the
transforming from voltage to velocity/acceleration not ac-
curate enough.
4) Distortion. The distortion of the sensor also has some in-
fluence on the data.

Conclusions
The sensor is just a device to track the shaking of the ground
accurately and the accuracy of its measurement depends
on two factors: EM performance and the coupling effect.
EM performance, such as dynamic range, sensitivity and
harmonic distortion, are important, but ultrahigh EM per-
formance cannot assure a progress of profile because of the
strong noise, the limited ability on noise attenuating and the
differences among the signals involved in one stack.
For example, an accelerometer’s harmonic distortion can
reach –90 dB (20dx: –60dB). However, reflection seismology
has many phases, from field acquisition to final application,
and many factors along the way can produce differences in
the results. Any factor which may produce a difference great
than –60 dB will obscure the advantage resulting from the
DSU3’s lower harmonic distortion (–90 dB). In fact, many
factors including noise, sources, propagation mediums, cou-
pling, height difference, and processing algorithms) can result
in differences greater than –60 dB.
So, currently, the accelerometer will not be able to show any
improvements, except under strictly controlled conditions.

Figure 9. Geometry of test 2.

October 2013 The Leading Edge 1209

 ACQUISITION/PROCESSING
Downloaded 11/11/13 to 130.238.140.107. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Figure 10. Comparison of two kinds of sensors in acceleration domain (middle offset).

No sensor can elevate the signal’s frequency, and the cause References
of the high-frequency signal attenuation is array effect or cou- Hauer, G., M. Hons, R. Stewart, D. Lawton, and M. Bertram, 2008,
pling effect. Field data comparison: 3C–2D data acquisition with geophones
If a geophone and an accelerometer are a few centimeters and accelerometers: 78th Annual International Meeting, SEG, Ex-
apart, what determines the signal frequency: accelerometer, panded Abstracts, http://dx.doi.org/10.1190/1.3054783.
Hons, M., R. Stewart, D. Lawton, and M. Bertram, 2007, Ground
geophone, or ground? Certainly the ground and the coupling
motion through geophones and MEMS accelerometers: Sensor
effect between ground and sensor, not the accelerometer or comparison in theory, modeling, and field data: 77th Annual
the geophone. The function of sensor is to detect the shaking International Meeting, SEG, Expanded Abstracts, http://dx.doi.
of the ground with how much, more or less, the electric noise org/10.1190/1.2792372.
and distortion there is, not to elevate the signal’s frequency. Maxwell, P. W. and B. Cain, 1999, Field test of a micromachined,
Single-receiving is not the patent of MEMS. electromechanical, digital seismic sensor: 69th Annual Inter-
Some comparisons were done between new data acquired national Meeting, SEG, Expanded Abstracts, http://dx.doi.
by accelerometers with new geometries, and improved process- org/10.1190/1.1821099.
ing measures etc. and older data acquired by geophones with Mougenot, D., 2004, Land seismic: Needs and answers: First Break,
old geometries and less sophisticated processing. Conclusions on 22, 59–63.
sensors are interfered by so many other factors. Sensor is only Mougenot, D. and N. Thorburn, 2004, MEMS-based accelerometers:
expectations and practical achievements: The Leading Edge, 23,
one of the possible reasons for the profile’s achievement. In ad-
246–250. http://dx.doi.org/10.1190/1.1690897.
dition, the correction on velocity/acceleration type and different Mougenot, D. and G. Volkov, 2004, Evaluation of digital accelerom-
frequency response should be done before a comparison. eter performance: 66th EAGE Conference and Exhibition.
The coupling ability of the sensor could affect the high-fre- Mougenot, D., 2003, How digital sensors compare to geophones?: Pre-
quency-end and the signal-to-noise ratio. Attempting to achieve sented at CPS/SEG International geophysical meeting (Beijing,2003).
better coupling effect, including lighter mass, smaller size, more Ronen, S., J. Gibson, R. Burnett, J. Roy, B. Montgomery, R. Kendall, L.
suitable shape and material for spiking etc., has equal importance Comeux, and H. Watt, 2005, Comparison of multicomponent data from
in the pursuit of better electric performance in the field. different MEMS sensors: 67th Meeting and Convention, EAGE, B021.
What kind of sensor can meet our geophysical require- Stotter, C. E., Angerer, and E. Herndler, 2008, Comparison of sin-
ments in seismic survey? gle sensor 3C MEMS and conventional geophone arrays for deep
The sensor with outstanding electric performance, good target exploration: 78th Annual International Meeting, SEG, Ex-
panded Abstracts, http://dx.doi.org/10.1190/1.3054782.
coupling effect, stability, and is economical and durable is our
target. Corresponding author: [email protected]

1210 The Leading Edge October 2013